ProfitPress Mega CDROM2 …eeware (MSDOS)(1992)(Eng)/ProfitPress-MegaCDROM2.B6I/UTILITY/DISK/DOGREVU.ZIP/DOGREVU.SCR

Text File | 1991-03-18 | 7.4 KB | 142 lines

                         Hard disk organization
                                 with
                             Disk OrGanizer


    When I added a hard disk to my PC, I was wasn't concerned with the
placement, on the disk, of the various programs that were accessed by
CONFIG.SYS and AUTOEXEC.BAT. I knew from my floppy days that boot-up could
be slow, indeed, if the drive had to seek far into the diskette for each
program, but hard disks are SO fast (68 milliseconds -- practically
instantaneous!) -- it didn't matter anymore.

    Soon, the disk was half full, and when I changed the boot-up stuff
(always tinkering), the newer versions were far up on the disk. And I began
to realize that there was a lot of thrashing going on! It was especially
noticeable when loading large programs, which, I realized, were badly
fragmented across the disk.

    Now, There are two important "asides" that should be brought up here --
ways to help smooth out the hard disk, things that are useful in any case.

FASTOPEN

    FASTOPEN is part of DOS; I think it was new in release 3.3. It speeds
thing up considerably by keeping an in-memory copy of the directories on the
hard disk. So after reading a file from (say) the root segment, all or most
of the root segment directory is kept in storage. So, next time AUTOEXEC
goes after a file, DOS does not have to re-read the directory! Provides a
very nice speedup; there is a dramatic drop in noise as AUTOEXEC loads
FASTOPEN.

IBMCACHE

    IBMCACHE, part of DOS at 4.0, I think, retains hard disk data in storage.
Much more than FASTOPEN: the latter keeps only directory information;
IBMCACHE buffers everything read from the disk! Now, in order to do any
good, you will need to have a large chunk of storage, say a few hundred
kilobytes, so it isn't practical unless you have some expanded or extended
memory available. I have 576K of my EMS (expanded) memory allocated to the
disk cache. (Of course, I no longer use FASTOPEN.) There is another nicety:
IBMCACHE is loaded from CONFIG.SYS, so the disk quiets down much sooner than
it did with FASTOPEN.

    It's nice, too, that often-used utility programs need to load from the
disk only once (or after long non-use), as, after the first time, it loads
from cache memory. I used to put frequently-used programs on my RAM disk,
but that's no longer necessary.

Back to the topic

    Of course, I knew from experience with floppies that my programs were
stored in fragments, and I'd read enough PC magazines to have seen
advertisements for disk de-fragmenting programs. They loudly proclaim how
fast your disk will go! Then one of the mags did a comparison of about a
dozen of them, giving very high marks to one called DOG, the Disk Organizer.
Best of all, it was a shareware program! My cheap little heart leapt! Only
$20!

    I found a version of DOG on JDR Electronic's bulletin board (408
559-0253), in the Hard Drive Utilities section; it's also on many others
as DOG206.ZIP. Soon, I was reading its instructions. I practiced by
operating on my RAM disk. I also tried some small files on a floppy.

    DOG permits you to specify the order in which files are to be placed on
the disk. Files that are not explicitly placed are put following the
explicit ones. In addition, you may place specific files at the TOP of your
disk! This is useful if you have a few very large files (such as data bases)
-- by putting them at the top, DOS doesn't have to seek past them looking for
other files. Of course, if the files are being lengthened, it is less
practical to put them high on the disk.

    All these file names are contained in a DOG control file called
ORDER.DOG, in the root segment of the disk. In addition to the above, you
may specify which directories are to be truncated; files that are not to be
moved (usually involved with copy-crippled programs); and a couple options
regarding presence of multi-processing and copy-protection.

    I listed first the things called from CONFIG and AUTOEXEC, then the
sub-directories. After the [HIGH] command, I listed the database files that
were to go at the top of the disk. Oh, yes, first of all were IBMBIO.COM and
IBMDOS.COM!

    The first time, of course, the poor thing had to move hundreds of
allocation units! But it's fast, and this took maybe 10 minutes. I noticed
that it was running very smoothly, with little to-and-fro seeking. ARRRG! I
had not disabled IBMCACHE!! I decided that the damage, if any, had already
been done, so I waited. DOG completed its run, re-checked its results, and
told me that there were zero fragmented files and one free area. OK so far.
I ran FLUSH to purge the disk cache, then CHKDSK. All was well! (I rather
suspect that FASTOPEN would have trouble, and would require a re-boot
afterward.)

    DOG has another very important feature: it has five options as to the
degree of rearranging to be done. You may specify a default in the ORDER
file, and may override it when calling DOG. The options are:

PACK -- defragment all files, pack them down leaving no unallocated space
    among them. Files stay in original order.

FILL -- defragment and move enough files to combine all unallocated space.
    This is almost always faster than PACK, but the files will be out of
    order, if that were to matter.

FAST -- do just enough rearranging to defragment all files. This option also
    does a little FILLing, so it's a very good one to use often. It may
    leave some free space among the files.

DIR -- defragment and arrange the (unspecified) files in the order that they
    appear in the directory. I don't know just how this works for files in
    various sub-directories.

DATE -- defragment and place oldest files first. This might reduce future
    file movement by migrating changing files up nearer the free space.

IN EACH CASE, files that are explicitly placed will also be moved as
necessary, and directories will be compressed. The program tests the
directories and the FAT, produces a little report giving the number of files,
fragments, and number of allocation units to be moved; then it asks whether
to continue. Upon a "Y" reply, it does its work, giving only the number of
allocation units already processed. It'd be nice if it gave the name of the
file being worked on, but it doesn't. Lastly, it again tests the FAT, etc.,
and gives the report again. You can see the results, remaining holes, if
any. At any time during the process, you many hit ESC to tell it to stop.
It will leave the disk in usable form, though it may be more fragmented than
before, and you may need to use CHKDSK to free some unallocated space. I've
never interrupted it, so I'm simply quoting the instructions here.

    In conclusion, DOG is a very flexible, convenient way to manage your hard
disk. It permits explicit ordering of files and directories, and offers a
choice of the degree of packing to be done. I run it nearly every time I've
modified files on my disk. I sometimes use it on floppy disks, but it's
really rather slow and noisy. I wish DOG had an option to bypass the
de-fragmentation, so that on floppies, I could get just the explicit files
and the directory packing. One can do this by explicitly listing all files
to be left alone, but that's seldom practical, especially since DOG does not
give the names of the fragmented files.

Clear Skies,
Jim Van Nuland
1990 February 10